AI chatbot disclosure generator
Tell people they're talking to AI at the start, in plain words, and tell them how to reach a person. This tool writes that notice for a website chat assistant, AI phone answering and AI-made content: a chat banner, a first message, a phone greeting, content labels, a "How we use AI" section and a privacy-policy paragraph, in English or Korean, with the rules behind it for the EU, UK, US and South Korea. Free, and it runs in your browser: nothing you type is sent or stored.
Add your business name to start.
This is a starting point, not legal advice. It reflects the rules listed above as we read them on 19 November 2026. Laws change; check the ones that apply to you.
OOO Chat is the website chat assistant OOO Works is preparing: it answers visitors only from your own website. It isn't available to try yet.
Watch your site weekly, free: we re-run our free checks every week and email you only when something changes. Nothing is for sale yet.
What a good AI disclosure says
- It's AI, not a person. Say it in plain words: "AI assistant", "not a person".
- At the start. Before or with the first message, or the first thing said on a call. The EU AI Act asks for it at the latest at the first interaction; Utah's safe harbor asks for it at the outset and throughout.
- It can be wrong. Ask people to check prices, availability and anything important.
- How to reach a person. An email, a phone number, or a "talk to a person" option.
- What happens to their words. Covered in your privacy policy: who processes the chat and how long you keep it.
Keep the banner visible for the whole chat, not only in the first message, and never give the assistant a human backstory or photo that suggests it's a person.
The rules, by region
Checked against the official texts on 19 November 2026. Laws change; follow the links for the current version.
European Union
- EU AI Act, Article 50(1) and 50(5) (From 2 August 2026): AI systems that interact directly with people must be designed so people are told they're interacting with an AI system, unless that's obvious. The duty falls on the provider (whoever builds the system), and the information must be clear and given at the latest at the first interaction.
- EU AI Act, Article 50(4) (From 2 August 2026): Businesses that publish AI-made deepfake images, audio or video must disclose it. AI-generated text published to inform the public on matters of public interest must be disclosed too, unless a person reviewed it and someone holds editorial responsibility.
- Digital Omnibus on AI, Regulation (EU) 2026/1744 (In force 27 July 2026): Kept Article 50's date. Providers of generative AI already on the market before 2 August 2026 have until 2 December 2026 for machine-readable marking of AI output (Article 50(2)).
- Commission guidelines on Article 50 (Published July 2026): The European Commission's guidance on how providers and deployers meet the transparency duties. Not binding law.
- GDPR, Articles 13 and 14 (In force): If the assistant handles personal data, your privacy information must say why, how long you keep it and who you share it with.
United Kingdom
- UK GDPR transparency (ICO guidance on AI) (In force): If an AI system processes personal data, your privacy information must give your purposes, how long you keep the data and who you share it with, at the time you collect it.
- UK GDPR, Articles 22A to 22D (From 5 February 2026): Changed by the Data (Use and Access) Act 2025. Significant decisions made solely by automated means need safeguards: telling people, letting them make representations, get human intervention and contest the decision.
- UK government approach to AI regulation (Policy): Sets non-statutory principles for regulators, including appropriate transparency. We found no UK law that specifically requires a chatbot to say it's AI.
United States
- California Business and Professions Code 17940 to 17943 (SB 1001) (From 1 July 2019): Using a bot online to mislead a person in California about its artificial identity, to encourage a sale or influence a vote, is unlawful. Disclosing that it's a bot, clearly and conspicuously, avoids liability.
- Utah Code 13-77-103 and 13-77-104 (From 7 May 2025): A business using generative AI with consumers must say it's AI and not a human if a customer clearly asks; regulated occupations must disclose up front in high-risk interactions. Disclosing at the outset and throughout the interaction is a safe harbor.
- Maine, 10 M.R.S. 1500-DD (From 24 September 2025): Using an AI chatbot in trade with a consumer in a way that may make a reasonable consumer think they're dealing with a human is an unfair trade practice, unless the consumer is clearly and conspicuously told they are not.
- FTC: Operation AI Comply (Since September 2024): "Using AI tools to trick, mislead, or defraud people is illegal." Federal consumer protection law applies to AI the same as anything else, including claims about what your AI can do.
- FCC ruling on AI voices in calls (FCC 24-17) (Since February 2024): AI-generated voices count as an "artificial voice" under the TCPA, so calls your AI places to people need their prior express consent. This covers calls you make, not calls you answer.
South Korea
- AI Basic Act, Article 31(1) (인공지능기본법 제31조) (From 22 January 2026): An AI business operator offering a product or service that uses generative or high-impact AI must tell users in advance that it runs on AI. A business that uses another company's AI to provide its own AI service counts as an operator.
- AI Basic Act, Article 31(2) and (3) (From 22 January 2026): Output of generative AI must be labelled as AI-generated, and sound, images or video that are hard to tell from real must be labelled so users clearly recognise them.
- Enforcement Decree, Article 23 (시행령 제23조) (From 22 January 2026): Notice can be given in the terms of use, on the user's screen, or posted where the service is provided. No notice is needed where the use of generative AI is obvious or the AI is only used internally.
- AI Basic Act, Article 43 (From 22 January 2026): Failing to give the advance notice in Article 31(1) can bring a fine of up to 30 million won. The government has said it will hold off fines for at least a year while businesses adjust.
- Personal Information Protection Act (In force): If the chat collects personal information, your privacy policy must cover it, including any outsourcing to an AI provider and transfers abroad.
Questions
Do I have to tell customers they're talking to a chatbot?
It depends on where your customers are. In the EU, Article 50 of the AI Act requires it from 2 August 2026. In South Korea, the AI Basic Act has required advance notice since 22 January 2026. In the US, California requires it when a bot is used to sell, Maine whenever a chatbot could make someone think it's a person, and Utah when a customer asks. We found no UK law that specifically requires it. Even where it isn't required, saying so up front is the safe choice and what customers expect.
When does the EU AI Act's chatbot rule start?
2 August 2026. The Digital Omnibus on AI (Regulation (EU) 2026/1744), in force since 27 July 2026, did not move that date for telling people they're talking to AI. It gave providers of generative AI already on the market until 2 December 2026 for machine-readable marking of AI output.
What should a chatbot disclosure say?
That the person is talking to an AI assistant, not a person; that answers can be wrong, so important details should be checked; and how to reach a person. Show it at the start, before or with the first message, and keep it visible during the chat. Utah's safe harbor, for example, asks for disclosure at the outset and throughout the interaction.
Do I have to label AI-written blog posts?
Usually not by law for an ordinary business blog in the EU: Article 50(4) covers AI-generated text published to inform the public on matters of public interest, and not where a person reviewed it and someone holds editorial responsibility. Korea's AI Basic Act asks AI business operators to label generative AI output, with exceptions where it's obvious. Many businesses label AI-assisted writing anyway, and we never let a label claim a person wrote something they didn't.
Do I need to tell callers that AI is answering the phone?
The same rules apply to a voice assistant as to a chat assistant: in the EU from 2 August 2026, in Korea since 22 January 2026, and in Utah the disclosure for regulated occupations must be given verbally at the start of a spoken interaction. Recording or transcribing calls has its own consent rules in many places, so check those too. In the US, calls your AI places to people need prior express consent under the FCC's 2024 ruling.
Is this legal advice?
No. It's starting text based on the rules listed on this page as we read them on 19 November 2026. Laws change and details matter, so check the rules that apply to you, and ask a lawyer if you work in a regulated field.
Is what I type stored anywhere?
No. The text is written in your browser by the script on this page. Nothing you type is sent to us or saved.
More free tools
- AI use policy generator: a plain-language policy for how your staff use AI tools.
- AI answer check: does ChatGPT recommend your business when customers ask?
- OOO Chat: a website chat assistant that answers only from your own website (in preparation).
Who made this
OOO Works, which is building done-for-you services for small businesses worldwide, including a website chat assistant. OOO Works is opening soon; nothing is on sale yet. We're not lawyers, and this tool doesn't give legal advice. About OOO Works.